In a system wherein a mainframe accesses input/output devices in accordance with configuration definitions of the input/output devices that are registered beforehand, the present invention relates to a technique for taking a plurality of generations of snapshots in a flexible manner without altering the configuration definitions of the input/output devices.
In general, when a mainframe performs input/output of data in respect of input/output devices such as disk volumes or tape volumes, it performs this input/output of data in accordance with the configuration of the input/output device that is registered in the operating system (OS). The system administrator must therefore perform beforehand the task referred to as configuration definition of the input/output devices. Typically, configuration definition of the input/output devices involves definition of the bus whereby the device performs input/output, definition of the number of the control device of the input/output device and definition of the storage device address. This makes it possible for these input/output devices to be used from the mainframe by putting these input/output devices into a condition in which they are logically incorporated with the mainframe to constitute the system.
Also, an upper limit on the number of input/output devices i.e. the number of device addresses that can be defined is laid down, depending on the type of mainframe and OS. Mainframes therefore cannot use a number of input/output devices that exceeds this upper limiting value.
Also, the amount of data that needs to be saved continues to increase with increase in the amounts of data that are accumulated. In many cases, the practice is adopted that data that are constantly accessed by a client business are saved on a disk array system and after the business that makes use of these data has finished with them, a back-up of the data is saved on a tape system. With this practice, if there should be any corruption of the data stored on the disk array system due for example to a mistake in operation, the data can be restored to the original condition by using the back-up acquired by the tape system. However, the time required for acquiring this back-up is proportional to the amount of data.
Furthermore, in recent years, with increases in the amount of data and increased penetration of 24 hour continuous service, cases have started to arise in which sufficient back-up time to back-up the data of the disk array system to a tape system cannot be secured. Cases in which use is made of so-called “snapshots” as a technique for maintaining a back-up of the data in question at a given time-point in a disk array system have therefore become more frequent. A “snapshot” as referred to here indicates a logical copy for restoring the data condition at a given time-point. If the practice is adopted of taking snapshots at as high a frequency as possible i.e. with time intervals that are short as possible, if it should subsequently be necessary to restore the data, the restoration time can be shortened by employing a snapshot that is closer to the time-point to which restoration is desired.